A Study on Adaptive Power Split Strategy of HEV Using Nonlinear System Identification



SAE WCX Digital Summit
Authors Abstract
In this paper, an adaptive power split method is proposed and studied. A 48V mild hybrid vehicle model is used as an example to illustrate the adaptive power split method. First, the dynamic programming (DP) is used to find the global optimal fuel consumption and corresponding internal combustion engine (ICE) and electrical motor (EM) torque for the different random driving cycles, which are used to identify the power split model. After the optimal ICE and EM power split trajectories are obtained, the correlation between the output dataset of optimal power split for EM and the input dataset of the battery SoC, powertrain torque demand and engine speed are analysed. The regressors, which are terms of linear and nonlinear combinations of input and output parameters, are selected according to the correlation analysis. A nonlinear Auto Regressive eXogenous (ARX) model, which is structured by the selected regressors, is determined for the adaptive power split policy through the nonlinear system identification from the input and output dataset mentioned above. Finally, the obtained adaptive power split model is used for simulation in a different driving cycle. The fuel consumption of the adaptive power split method, rule-based strategy and dynamic programming (DP) strategy which is optimised against the certain driving cycle are compared. Although the adaptive power split is obtained based on the different vehicle driving cycle information, the fuel consumption of adaptive power split method is still slightly better than the rule-based strategy. The robustness and performance of the proposed method using nonlinear ARX model is proved in a certain degree.
Meta TagsDetails
Zhao, S., "A Study on Adaptive Power Split Strategy of HEV Using Nonlinear System Identification," SAE Technical Paper 2021-01-0779, 2021, https://doi.org/10.4271/2021-01-0779.
Additional Details
Apr 6, 2021
Product Code
Content Type
Technical Paper